Hard soils—soft rocks: modelling the soil behaviour—selection of soil parameters, general report PDF

Preview

Summary

General report: Modelling the soil behaviour - Selection of soil parameters Michael J. Kavvadas National Technical University of Athens, Greece ABSTRACT: The engineering characteristics of Hard Soils - Soft Rocks (HSSR) and their relationships to the properties of soft natural soils and hard rocks a...

Description

Accelerat ing t he world's research.

Hard soils—so rocks: modelling the soil behaviour—selection of soil parameters, general report Michael Kavvadas

Related papers

Download a PDF Pack of t he best relat ed papers 

Charact erisat ion and engineering propert ies of t he Corint h marl Michael Bardanis, Michael Kavvadas

An anisot ropic Model for St ruct ured Soils:: Part I: T heory Michael Kavvadas, Giorgos Belokas Invest igat ing t he influence of microst ruct ure, loading hyst ory and fissuring on t he clay response Claudia Vit one

General report: Modelling the soil behaviour - Selection of soil parameters Michael J. Kavvadas National Technical University of Athens, Greece

ABSTRACT: The engineering characteristics of Hard Soils - Soft Rocks (HSSR) and their relationships to the properties of soft natural soils and hard rocks are investigated, considering soil structure as the governing parameter. Modelling HSSR is thus reduced to modelling soil structure, in a way which unifies the effects of stress-history, diagenetic bonding and matrix suction and can include all materials requiring a soil mechanics approach. The paper describes a conceptual and analytical framework for modelling soil structure as an extension of classical plasticity-based models for reconstituted soils. Several existing models for HSSR are described and interpreted with reference to this framework. Selection of soil parameters is linked to modelling, as such procedures are performed against the background of a constitutive model. The difficulties and uncertainties in selecting appropriate values for soil parameters are discussed via characteristic examples. 1 INTRODUCTION For centuries, geotechnical design was based on "experience" and "judgement", i.e., on the ability of engineers to identify relevant past experience and apply it in similar circumstances. This approach is now supplemented with quantitative analytical methods, especially since numerical techniques have become commonplace in the analysis of geotechnical problems. The use of analytical methods in design typically requires the following procedure: 1. Adoption of an analytical model for the problem under investigation, i.e., a model of the geometry, boundary conditions, initial conditions, loading sequence, etc. 2. Characterisation of the materials involved and investigation of their engineering properties. 3. Adoption of appropriate constitutive models for the materials of interest and selection of the parameters required by the models. 4. Numerical analyses using the above geometrical and constitutive models. The above methodology may require iterative updating of the design assumptions and is often combined with validation procedures based on measured field performance. Session II of this Symposium is related to item 3 above, when the materials of interest belong to the class of "Hard M. Kavvadas, General report - Session II

Soils - Soft Rocks" (HSSR). HSSR have been traditionally viewed as borderline cases between soils and rocks (e.g. Morgenstern and Eigenbrod, 1974). Being fringe dwellers, they have not really had the attention that their frequency of occurrence world-wide appears to demand. This may be partly due to the fact that, comparable to soft soils, HSSR are engineered quite successfully on the basis of experience-judgement. This may be attributed to their high stiffness and strength, normally exceeding the design requirements in typical projects. Despite that, HSSR have occasionally caused catastrophic failures, such as the famous 1914-15 Culebra slides along the Panama Canal. These slides were associated with the Cucaracha formation, a highly slickensided montmorillonite-rich clay-shale (Lutton and Banks, 1970). In the last years, it has become urgent to improve our understanding of the mechanical behaviour of HSSR, as modern large projects often place very strict requirements on the control of ground movements. It has thus become necessary to identify the characteristics of HSSR that distinguish them from other geomaterials. Once an acceptable level of understanding is achieved, conceptual frameworks of behaviour are formulated and implemented in mathematical (constitutive) models. 1

This state-of-the-art report follows the above logical sequence: it presents the main engineering characteristics of HSSR and describes the development of conceptual and analytical models of behaviour. The second topic of the report (selection of soil parameters) is intimately linked to modelling, as soil parameters are always selected against the background of a constitutive model. Thus, parameter selection procedures are related not only to the accuracy of experimental measurements but also to their relevance with the constitutive model to be used in engineering analysis. 2 ROCK OR SOIL MECHANICS APPROACH ? A fundamental question in studying borderline materials such as HSSR is related to the required approach, i.e., whether Soil Mechanics or Rock Mechanics (or maybe something else) is most appropriate. The answer to this question is crucial since, despite arguments in favour of pursuing unified approaches for all geomaterials (e.g. Johnston and Novello, 1994), Soil Mechanics and Rock Mechanics still use different and often contrasting methods of investigation, classification, testing, interpretation, modelling and design. Furthermore, the answer to the above question is closely linked with the classification of HSSR, i.e., the determination of the materials included in the class of HSSR. Traditionally, the transition from soil to rock is based on the magnitude of the unconfined strength of the intact material (e.g. Morgenstern and Eigenbrod, 1974). In this manner, HSSR include a range of geomaterials with strengths between those of "hard" (or "indurated") soils and "soft" (or "weak") rocks, e.g. between 0.3 MPa and 5 MPa according to the British Standards (BS5930:1981). While such classifications are invaluable in creating and/or settling contractual claims related to the nature of the encountered materials, from the aspect of modelling it is preferable to set the lower and upper bounds of HSSR based on similarities of engineering behaviour rather than the value of a single property (i.e. strength). In the following sections, we argue that HSSR have similar patterns of behaviour with natural "soft" soils (which are studied using soil mechanics approaches). Thus, for the purposes of modelling, the lower bound of the spectrum of HSSR is not a major issue for discussion. Using the above principle of "similarity in behaviour", the upper bound of HSSR has to be placed at a position where Soil M. Kavvadas, General report - Session II

Mechanics becomes irrelevant and Rock Mechanics approaches are required. Rock Mechanics is sometimes considered as the science studying the brittle and truly-cohesive rocks, while Soil Mechanics studies the plastic and purely frictional soils. This is certainly false, since such a distinction between soils and rocks is very ambiguous; in fact, rocks can deform plastically (like soils) at high confinement, overconsolidated soils are brittle (like rocks) at low confinement, and most natural "soft" soils possess true cohesion (like rocks) due to inter-particle bonding (e.g. Elliot and Brown, 1985; Leroueil and Vaughan, 1990). In the present report, Rock Mechanics is considered as an approach to study geomaterials satisfying both of the following criteria: 1. They have macro-structural features (i.e., largescale discontinuities such as bedding planes, joints, faults and the like) and the strength along these discontinuities (due to friction, interlocking, etc) is appreciably lower than the strength of the intact material. In such cases, behaviour is governed mainly by the discontinuities as they are the weak link of the system (e.g. Clayton and Serratrice, 1993). 2. Excess pore pressures have a negligible effect on their response. This condition is satisfied in problems where the loading rate is sufficiently slow to allow for practically complete drainage, i.e., for materials having sufficiently low compressibility (e.g. "hard" rocks) or sufficiently high permeability to either develop negligible excess pore pressures or dissipate them quickly. On the contrary, a Soil Mechanics approach is required if either of the above criteria is not satisfied. Thus, Soil Mechanics is an approach for geomaterials without macro-structural features (or, if such features are present, they have negligible effects), or for problems where excess pore pressures are important. According to the above postulate, a rock mechanics approach is not always required for problems involving "hard rocks", since the same material, depending on the circumstances (mainly geometrical scale and stress level), may require either a soil mechanics or a rock mechanics approach. Examples of hard rocks requiring a soil mechanics approach are: 1. Very deep tunnels where, due to the high confinement and the closure of the joints, the frictional strength of the discontinuities may become comparable to the intact strength even of hard rocks (Hyett and Hudson, 1990). In such a 2

case, the rock-mass deforms and fails plastically as a practically homogeneous material thus making soil mechanics approaches more appropriate. On the contrary, a shallow tunnel in the same hard rock in most cases would require a rock mechanics approach, since the response of the rock-mass would be governed by the strength along the discontinuities (as the low confinement has a disproportionally high deleterious effect on the frictional strength of the discontinuities compared to the mostly cohesive strength of the intact rock). 2. Deep oil-bearing sandstones, where the compressibility of the formation becomes important (due to appreciable de-pressurisation of the pore fluid) and oil extraction is controlled by soil mechanics consolidation theories. At the other extreme, a typical example of a "soil" problem requiring a Rock Mechanics approach is the development of slides in some stiff highly fissured ("scaly") Italian clay-shales under drained conditions. In these materials, although large-scale discontinuities are usually initially absent, shear failure is controlled by slippage along closely-spaced and favourably oriented small-scale fissures which gradually become inter-connected (see Figure 1), causing a large-scale localisation of the deformation and eventually the development of a continuous large-scale shear surface (Picarelli and Olivares, 1998). Experimental evidence (Olivares et al, 1997; Picarelli and Olivares, 1998) indicates that the shear strength along the fissures is very low (because their surfaces are polished or even slickensided) and thus the strength of the large-scale discontinuities (formed by inter-connected fissures) is appreciably smaller than the strength of the "intact' material (the hard fragments bounded by the fissures) and even smaller than the post-rupture strength of the reconstituted material. The investigation of slides

Figure 1. Mechanism of deformation and rupture of intensely fissured ("scaly") Italian clay shales (after Olivares et al, 1997) M. Kavvadas, General report - Session II

under drained conditions in these materials may require a rock mechanics approach, since wedgetype potential failure planes along the prevailing orientation of the discontinuities are more appropriate than curved failure surfaces. Based on the above arguments, it is concluded that the upper bound of HSSR may be set where a Rock Mechanics approach becomes required. Equivalently, it may be stated that HSSR are: 1. Materials not influenced significantly by macrostructural features (i.e., large-scale discontinuities) for the specific problem circumstances, or 2. Materials in problems where excess pore pressures are important. Thus, by definition, HSSR can be studied using soil mechanics approaches. Excess pore water pressures often play a minor (or even negligible) role in problems involving HSSR. As the stiffness of the material increases, its compressibility decreases and becomes comparable, or even lower, than the compressibility of the pore water. In such cases excess pore pressures are negligible (i.e., Skempton's "B" parameter approaches zero) and their dissipation is rapid. Thus, the influence of excess pore pressures in HSSR decreases as the stiffness of the material increases, but it is not always true that HSSR are not influenced by excess pore pressures. The potential effect of macro-structural features on the behaviour of HSSR can be assessed by comparing the shear modulus obtained from field measurements of the shear wave velocity (which includes any effects of large-scale discontinuities) to the very-small-strain shear modulus (Gmax) measured

Figure 2. Comparison of very-small-strain shear moduli (Gmax) from triaxial tests and corresponding values (Gf) from seismic surveys in sedimentary soft rocks and cement-treated soils (after Tatsuoka and Kohata, 1994).

3

Figure 3. One-dimensional compression curves. Soft (su=20-30 kPa) slightly overconsolidated (OCR=1.3-1.5) Bothkennar clay (after Nash et al, 1992).

Figure 4. Anisotropic triaxial compression curves. Stiff heavily overconsolidated Vallericca clay (after Amorosi and Rampello, 1998).

on high quality undisturbed samples which do not include large-scale discontinuities (e.g. in resonant column, torsional shear or bender-element tests). Experimental evidence (e.g. Tatsuoka and Shibuya, 1992; Tatsuoka and Kohata, 1994) shows that the shear modulus obtained from field measurements of the shear wave velocity in soft rocks without discontinuities is very well comparable to the average very-small-strain shear modulus measured in well-controlled triaxial tests on high quality undisturbed samples. Thus, it is reasonable to conclude that any appreciable difference between field and laboratory values of Gmax is due to the effects of large-scale discontinuities, provided that sample disturbance is not important in the laboratory-measured values of Gmax; such measurements may be used as a criterion for the potential effects of macro-structural features, i.e., as a criterion for the applicability of soil mechanics approaches in the study of HSSR. Tatsuoka and Kohata (1994) report several field applications of the above procedure in Japanese soft rocks where the effects of widely spaced discontinuities on Gmax are negligible, an indication that the above assessment M. Kavvadas, General report - Session II

Figure 5. Isotropic compression curves. Fine grained Neapolitan tuff, qu=8MPa (after Aversa and Evangelista, 1998).

Figure 6. Isotropic compression curve. Hard Gosford sandstone, qu=80MPa (after Novello and Johnston, 1995)

method is a reliable criterion (see Figure 2). This may be due to the fact that HSSR usually have relatively low shear strength (e.g. less than 5-10 MPa) and few (if any) widely-spaced large-scale discontinuities. In most cases, such discontinuities have a minor effect on the mechanical behaviour of HSSR because the low strength of the intact material is not in high contrast to the strength of the discontinuities. Typical examples of materials behaving as HSSR in common engineering projects include hard overconsolidated clays and clay-shales, soft sedimentary rocks (e.g. mudstones, claystones, marls, shales, calcarenites and weak limestones), weak pyroclastic rocks (e.g. tuffs), cemented coarse-grained materials (e.g. weak sandstones) and even very weathered hard rocks and residual soils (since the high contrast of strengths between the intact material and the discontinuities is softened by weathering). The majority of these materials are geologically relatively young (Pliocene or younger), with some notable exceptions like the Jurassic Upper Lias clay in Britain, the upper Cretaceous argillaceous 4

Figure 7. Shear stress (top) and volumetric strain (bottom) versus axial strain plots. Drained triaxial tests at various confining pressures. St. Vallier soft sensitive Canadian clay (after Lefebvre, 1970).

formations of the Canadian interior plains, the Eocene London clay, etc. Because of their relatively young age and/or their geographical location, most of these materials have not been subjected to high tectonic stresses and thus include few or no largescale discontinuities. Furthermore, because of their relatively low stiffness, they tend to deform plastically and are not prone to the development of large-scale discontinuities, even when subjected to tectonic deformations under moderate and large geological pressures. It is the author's belief that in most cases involving materials traditionally described as "hard" (or indurated) soils and "soft" (or weak) rocks, the effects of large-scale discontinuities may be neglected and a soil mechanics approach can be used in the constitutive modelling of these materials. For all the above reasons, this report is restricted to the description of soil mechanics approaches in the conceptual and constitutive modelling of HSSR lacking significant macro-structural features. 3 EFFECTS OF STRUCTURE IN SOILS / HSSR By postulating that HSSR should be studied by soil mechanics approaches, an obvious question arises: "In what respect is the behaviour of HSSR different M. Kavvadas, General report - Session II

Figure 8. Shear stress (top) and volumetric strain (bottom) versus axial strain plots. Drained triaxial tests at various confining pressures. Fine grained Neapolitan tuff, qu=8MPa (after Aversa et al, 1993).

than the behaviour of soils traditionally studied by Soil Mechanics?" or, equivalently, "What are the required extensions of classical soil mechanics models in order to include HSSR?" This section argues that the mechanical characteristics of HSSR stem mainly from their "micro-structure" and thus, classical soil mechanics models must be extended to include the engineering effects of "micro-structure" (referred to as "structure" in the following presentation). Abundant experimental evidence supports that HSSR follow similar patterns of mechanical behaviour as natural "soft" soils and even intact "hard" rocks, in both compression and shear. To illustrate that, Figures 3-6 compare the compressive response of a soft slightly overconsolidated clay (Bothkennar), a stiff heavily overconsolidated clay (Vallericca), a soft rock (fine-grained Neapolitan tuff) and a hard rock (Gosford sandstone). The uniaxial compressive strengths of these materials differ by more than three orders of magnitude (40 kPa - 80 MPa). All compression curves show an initially stiff response, followed by a relatively distinct yield and a softer behaviour in the post-yield domain. Obviously, the stress level at yield varies 5

Figure 9. Shear stress (top) and volumetric strain (bottom) versus axial strain plots of triaxial tests at various confining pressures. High porosity oolitic limestone, qu=23 MPa (after Elliot and Brown, 1985).

significantly among the materials tested and ranges between 70-80 kPa for the soft Bothkennar clay and 200 MPa for the hard sandstone. Figures 7-9 illustrate analogous similarities in the shear stress and volumetric strain response of several materials during drained triaxial tests performed at various confining pressures. These materials include a soft sensitive Canadian clay (St. Vallier qu=0.1MPa), a soft rock (fine grained Neapolitan tuff, qu=8MPa) and a moderately hard rock (oolitic limestone, qu=23MPa). At low confinement, they are brittle and dilatant, they fail with the development of localised shear bands and they usually show appreciable strain-softening in the post-rupture domain. At high confining pressures, they deform in a plastic and contractant way, usually they fail without a peak strength and without the development of shear bands and they strain-harden appreciably with shear strain. Obviously, the magnitude of the confining pressure at the transition between brittle and ductile behaviour is very different, ranging between a few kPa for the soft clay and about 7 MPa for the limestone. The properties of intact hard-rocks are gov...